The Earliest Finish Time And Duplication Based Scheduling Algorithm In Grid Computing Systems

As a typical distributed system, the grid is composed of a large number of the shared and heterogeneous resources which provide a tremendous computing power. The resources of grid computing are wide-area distributed, self-management, heterogeneous, and dynamic load changes, which makes the tasks scheduling in the grid environment face much more complex problems than the tasks scheduling in the traditional distributed environment. The tasks scheduling in the grid is the core technology of the grid.During the tasks scheduling in grid, communication delay that affect scheduling is an important factor, which not only leads to waiting between tasks, but also make processing units own too much idle time interval. At the same time, how to make use of resources and how to select resources are also important factors. In this paper, the tasks scheduling strategy considers the communication delay and the impact of resources preprocessing.On the one hand, because the traditional grid tasks scheduling doesn't consider resources preprocessing which is based on resources characteristics, the scheduling don't have good use of resources. In this paper, using fuzzy clustering method for the resources preprocessing reduces the scheduling process time which is spent in selecting the processing unit.On the other hand, combination of list scheduling and task duplication scheduling, CFTD algorithm is presented, (Cluster Earliest Finish Time Duplication resource-based clustering strategy for the earliest completion time of replication). When there is a period of idle time on the processor, CFTD attempts to insert suitable immediate parent nodes of the current selected node, to reduce its waiting time on the processor, which is helpful to advancing the earliest starting time of this candidate tasks. The experimental results show that the algorithm proposed in this paper is better than HEFT algorithm.